Radio Map-assisted CSI Tracking with Uncertain Locations in Massive MIMO Networks

Massive multiple-input multiple-output (MIMO) systems offer significant potential to enhance wireless communication performance, yet efficient and accurate channel state information (CSI) tracking remains a key challenge, particularly in dynamic urban settings. To address this, we propose a radio mapassisted framework for CSI tracking and trajectory discovery, relying on sparse channel observations. The radio map is redefined as a mapping from spatial positions to deterministic channel covariance matrices, which captures the complex and time-varying characteristics of urban wireless environments. Leveraging these covariance maps, we develop a CSI tracking method that enables accurate estimation using only single-dimensional observations collected during user movement. Furthermore, we present an efficient algorithm that constructs and continuously refines the radio map through sequential sparse observations, even when location labels are uncertain. Numerical results based on real city maps and ray-tracing MIMO channel datasets show that the proposed framework significantly outperforms baseline methods in both accuracy and adaptability.